Hence, the increased rainfall may simply be due to increased moisture flux. Although seemingly different in nature, these senarios may be linked through variations in cross-equatorial heat transport. Hasenrath & Greischar (1993) calculated that about 70% of the water which precipitates over Indian during the southwest monsoon was evaporated from the Indian Ocean south of 10｡?. They assumed that the heat required to sustain vigorous evaporation in the winter hemisphere was provided by cross-equatorial transport of heat from the Arabian Sea. Little was known about the Indonesian throughflow at the time Hasenrath & Greischar made their analysis, but it is now recognised that its transport of about 20 Sv (1 Sv=1 x 106m3s-1) of warm western equatorial Pacific water during the southwest monsoon could be a significant factor in the overall heat budget for the Indian Ocean.
An important first step in researching interannual and decadal-scale variability in the Indian Ocean heat budget is to investigate the processes producing variability in the throughflow's heat and fresh-water input into the Indian Ocean. Models have shown that its net mass transport varies with phase of the Southern Oscillation. It is weak during EI Nino, and strong during La Nina. The amplitude of the interannual signal may be as large as 4 Sv. Its T, S properties are determined by whether it is fed by the cold, fresh Mindanao Current or warm, salty South Equatorial Current (SEC). Historical ship drift data indicate that the SEC wholly retroflects into the North Equatorial Countercurrent to the north of Halmahera, and part of the Mindanao Current enters the archipelago to make up the throughflow. The paths taken by the WEPOCS buoys, see Lukas et al. (1991), are in agreement with this general picture. Hydrographic data from the ARLlNDO cruises, Gordon & Fine (1996), show a more complex picture with water masses of the SEC present in the Halmahera, MaluKu and Seram Seas below 100m, and those of the Mindanao Current confined to Makassar Strait and the Maluku Sea.
In this manuscript, based on Wajsowicz (1998) and (1999a), the robustness to variations in the Pacific wind-stress curl of the above described closure of the wind-driven gyres in the western equatorial Pacific, and throughflow composition, is investigated. Although the implied variations in the heatand fresh-water input into the Indian Ocean may not be as large as those due to transport variations, and so their impact on the Indian Ocean heat budget not as great, knowledge of their existence is useful for two reasons. Firstly, feeding of the throughflow by the Mindanao Current implies a compensatory Pacific cross-equatorial mass transport, whereas feeding by the SEC does not require any interhemispheric exhange. Secondly, the throughflow's T, S properties determine its fate within the Indian Ocean.
